Implementation of 5*5 Wumpus game using Q-Learning algorithm with python and javascript
There are two files, one is the implementation without graphics with python, and the other is implemented graphically with javascript. I used simple Q-Learning formula. (see the equation below)
goal reward: 100
holes and wall rewards = -100
other states reward = -1
discount factor: 0.8
episodes = 500
the following animation shows the performance of the Q-Learning algorithm with 100 episodes for training: (you can also see the changes of Q(s, a) per episode)
